[TRTLLM-7008][fix] fix wideEP weights loading and args #6789
Conversation
Signed-off-by: Dongxu Yang <[email protected]>
📝 WalkthroughWalkthroughUpdated a kwarg name in an fp4_quantize call within Changes
Estimated code review effort🎯 2 (Simple) | ⏱️ ~8 minutes Suggested reviewers
✨ Finishing Touches
🧪 Generate unit tests
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
SupportNeed help? Create a ticket on our support page for assistance with any issues or questions. CodeRabbit Commands (Invoked using PR/Issue comments)Type Other keywords and placeholders
Status, Documentation and Community
|
|
/bot run |
|
PR_Github #14786 [ run ] triggered by Bot |
|
PR_Github #14786 [ run ] completed with state |
Signed-off-by: Dongxu Yang <[email protected]>
|
/bot run |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 0
🔭 Outside diff range comments (2)
tensorrt_llm/_torch/modules/fused_moe/quantization.py (2)
1580-1599: Gate CPU execution in fused MoE quantization methods
The CPU path still calls CUDA-only ops (block_scale_interleave+torch.cuda.synchronize), which will fail whendevice.type!="cuda". We need to skip the interleave/sync on CPU in both methods and ensure it runs later when scales land on CUDA.• In
load_expert_w3_w1_weight_scale_nvfp4(around lines 1563–1576)
• Inload_expert_w2_weight_scale_nvfp4(around lines 1589–1599)Suggested patch for each location:
orig_shape = dst_…_weight_scale.shape - dst_…_weight_scale_interleaved = torch.ops.trtllm.block_scale_interleave( - dst_…_weight_scale.view(float4_sf_dtype) - ).view(self.block_scales_dtype).reshape(orig_shape) - - torch.cuda.synchronize() - - dst_…_weight_scale.copy_(dst_…_weight_scale_interleaved) + # Only interleave & sync on CUDA + if device.type == "cuda": + interleaved = torch.ops.trtllm.block_scale_interleave( + dst_…_weight_scale.view(float4_sf_dtype) + ).view(self.block_scales_dtype).reshape(orig_shape) + torch.cuda.synchronize() + dst_…_weight_scale.copy_(interleaved)• After patching, verify that when CPU-staged scales are later transferred to CUDA (e.g. by the layer load balancer/weight sharer),
block_scale_interleaveis applied before final use.
1540-1576: Only guard CUDA synchronization in CPU staging, not the interleave callThe CPU code path already supports
block_scale_interleave, so we shouldn’t skip swizzling on CPU—only the GPU‐only sync needs guarding. Replace the unguardedtorch.cuda.synchronize()with a device check:--- a/tensorrt_llm/_torch/modules/fused_moe/quantization.py +++ b/tensorrt_llm/_torch/modules/fused_moe/quantization.py @@ lines 1567-1576 - dst_w3_w1_weight_scale_interleaved = torch.ops.trtllm.block_scale_interleave( - dst_w3_w1_weight_scale.view(float4_sf_dtype) - ).view(self.block_scales_dtype).reshape(orig_shape) - - torch.cuda.synchronize() - - dst_w3_w1_weight_scale.copy_(dst_w3_w1_weight_scale_interleaved) + # Swizzle block scales for both CPU and CUDA paths + dst_w3_w1_weight_scale_interleaved = torch.ops.trtllm.block_scale_interleave( + dst_w3_w1_weight_scale.view(float4_sf_dtype) + ).view(self.block_scales_dtype).reshape(orig_shape) + + # Only synchronize on CUDA devices + if device.type == "cuda": + torch.cuda.synchronize() + + dst_w3_w1_weight_scale.copy_(dst_w3_w1_weight_scale_interleaved)No further CPU‐to‐CUDA “fix-up” hook is needed: the initial load already performs the interleave on the host, and the registered migrate‐weight functions don’t re‐swizzle. This change ensures CPU staging won’t error on
torch.cuda.synchronize()while preserving correct block‐scale interleaving everywhere.
🧹 Nitpick comments (2)
tensorrt_llm/_torch/modules/fused_moe/quantization.py (2)
350-361: Comment grammar + clarity: make the intent explicitThanks for moving the comment before the device capture; this addresses the earlier feedback. Nit: tighten the wording and reference the specific tensor for clarity.
Apply this diff:
- # device don't have to be 'cuda', e.g. 'cpu' for online EPLB + # Device may be CPU (e.g., for online EPLB); derive device from dst_w3_w1_weight. device = dst_w3_w1_weight.device
375-383: Comment grammar + clarity: make the intent explicitSame nit as above—tighten the wording and reference the destination tensor.
Apply this diff:
- # device don't have to be 'cuda', e.g. 'cpu' for online EPLB + # Device may be CPU (e.g., for online EPLB); derive device from dst_w2_weight. device = dst_w2_weight.device
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
tensorrt_llm/_torch/modules/fused_moe/quantization.py(4 hunks)
🧰 Additional context used
📓 Path-based instructions (2)
**/*.py
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
**/*.py: Python code should conform to Python 3.8+.
Indent Python code with 4 spaces. Do not use tabs.
Always maintain the namespace when importing in Python, even if only one class or function from a module is used.
Python filenames should use snake_case (e.g., some_file.py).
Python classes should use PascalCase (e.g., class SomeClass).
Python functions and methods should use snake_case (e.g., def my_awesome_function():).
Python local variables should use snake_case. Prefix k for variable names that start with a number (e.g., k_99th_percentile).
Python global variables should use upper snake_case and prefix G (e.g., G_MY_GLOBAL).
Python constants should use upper snake_case (e.g., MY_CONSTANT).
Avoid shadowing variables declared in an outer scope in Python.
Initialize all externally visible members of a Python class in the constructor.
For interfaces that may be used outside a Python file, prefer docstrings over comments.
Comments in Python should be reserved for code within a function, or interfaces that are local to a file.
Use Google style docstrings for Python classes and functions, which can be parsed by Sphinx.
Attributes and variables in Python can be documented inline; attribute docstrings will be rendered under the class docstring.
Avoid using reflection in Python when functionality can be easily achieved without it.
When using try-except blocks in Python, limit the except to the smallest set of errors possible.
When using try-except blocks to handle multiple possible variable types in Python, keep the body of the try as small as possible, using the else block to implement the logic.
Files:
tensorrt_llm/_torch/modules/fused_moe/quantization.py
**/*.{cpp,h,hpp,cc,cxx,cu,py}
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
All TensorRT-LLM Open Source Software code should contain an NVIDIA copyright header that includes the current year. This includes .cpp, .h, .cu, .py, and any other source files which are compiled or interpreted.
Files:
tensorrt_llm/_torch/modules/fused_moe/quantization.py
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: Pre-commit Check
|
/bot run |
|
PR_Github #14984 [ run ] triggered by Bot |
|
PR_Github #14984 [ run ] completed with state |
Summary by CodeRabbit
New Features
Refactor
Description
Test Coverage
GitHub Bot Help
/bot [-h] ['run', 'kill', 'skip', 'reuse-pipeline'] ...Provide a user friendly way for developers to interact with a Jenkins server.
Run
/bot [-h|--help]to print this help message.See details below for each supported subcommand.
run [--reuse-test (optional)pipeline-id --disable-fail-fast --skip-test --stage-list "A10-PyTorch-1, xxx" --gpu-type "A30, H100_PCIe" --test-backend "pytorch, cpp" --add-multi-gpu-test --only-multi-gpu-test --disable-multi-gpu-test --post-merge --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" --detailed-log --debug(experimental)]Launch build/test pipelines. All previously running jobs will be killed.
--reuse-test (optional)pipeline-id(OPTIONAL) : Allow the new pipeline to reuse build artifacts and skip successful test stages from a specified pipeline or the last pipeline if no pipeline-id is indicated. If the Git commit ID has changed, this option will be always ignored. The DEFAULT behavior of the bot is to reuse build artifacts and successful test results from the last pipeline.--disable-reuse-test(OPTIONAL) : Explicitly prevent the pipeline from reusing build artifacts and skipping successful test stages from a previous pipeline. Ensure that all builds and tests are run regardless of previous successes.--disable-fail-fast(OPTIONAL) : Disable fail fast on build/tests/infra failures.--skip-test(OPTIONAL) : Skip all test stages, but still run build stages, package stages and sanity check stages. Note: Does NOT update GitHub check status.--stage-list "A10-PyTorch-1, xxx"(OPTIONAL) : Only run the specified test stages. Examples: "A10-PyTorch-1, xxx". Note: Does NOT update GitHub check status.--gpu-type "A30, H100_PCIe"(OPTIONAL) : Only run the test stages on the specified GPU types. Examples: "A30, H100_PCIe". Note: Does NOT update GitHub check status.--test-backend "pytorch, cpp"(OPTIONAL) : Skip test stages which don't match the specified backends. Only support [pytorch, cpp, tensorrt, triton]. Examples: "pytorch, cpp" (does not run test stages with tensorrt or triton backend). Note: Does NOT update GitHub pipeline status.--only-multi-gpu-test(OPTIONAL) : Only run the multi-GPU tests. Note: Does NOT update GitHub check status.--disable-multi-gpu-test(OPTIONAL) : Disable the multi-GPU tests. Note: Does NOT update GitHub check status.--add-multi-gpu-test(OPTIONAL) : Force run the multi-GPU tests in addition to running L0 pre-merge pipeline.--post-merge(OPTIONAL) : Run the L0 post-merge pipeline instead of the ordinary L0 pre-merge pipeline.--extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx"(OPTIONAL) : Run the ordinary L0 pre-merge pipeline and specified test stages. Examples: --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx".--detailed-log(OPTIONAL) : Enable flushing out all logs to the Jenkins console. This will significantly increase the log volume and may slow down the job.--debug(OPTIONAL) : Experimental feature. Enable access to the CI container for debugging purpose. Note: Specify exactly one stage in thestage-listparameter to access the appropriate container environment. Note: Does NOT update GitHub check status.For guidance on mapping tests to stage names, see
docs/source/reference/ci-overview.mdand the
scripts/test_to_stage_mapping.pyhelper.kill
killKill all running builds associated with pull request.
skip
skip --comment COMMENTSkip testing for latest commit on pull request.
--comment "Reason for skipping build/test"is required. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.reuse-pipeline
reuse-pipelineReuse a previous pipeline to validate current commit. This action will also kill all currently running builds associated with the pull request. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.